\begin{table}[H] \centering
\newcolumntype{R}{>{\raggedleft\arraybackslash}X}
\newcolumntype{L}{>{\raggedright\arraybackslash}X}
\newcolumntype{C}{>{\centering\arraybackslash}X}

\caption{Heterogeneity analysis}
\label{tab:baseline_het_pres_parl_ballot}
\begin{tabularx}{\linewidth}{@{}lCCCCCCC@{}}

\toprule
 & & \multicolumn{2}{c}{\textbf{Election type}} & \multicolumn{2}{c}{\textbf{Incumbent on ballot}} & \multicolumn{2}{c}{\textbf{Checks \& balances}} \tabularnewline & (1) & (2) & (3) & (4) & (5) & (6) & (7) \tabularnewline 
{}&{Baseline}&{Pres.}&{Parl.}&{Yes}&{No}&{High}&{Low} \tabularnewline
\midrule \addlinespace[\belowrulesep]
\textbf{Economic performance}&0.269***&0.396***&0.200*&0.285**&0.218*&0.159&0.357** \tabularnewline
&(0.101)&(0.168)&(0.115)&(0.160)&(0.150)&(0.108)&(0.171) \tabularnewline
GDP p.c. growth&0.044&-0.020&0.087&-0.285&0.074&-0.030&0.120 \tabularnewline
&(0.155)&(0.239)&(0.200)&(0.242)&(0.265)&(0.152)&(0.243) \tabularnewline
(Minus) Inflation&0.430**&1.093***&0.166&0.425**&0.293&0.312*&0.530** \tabularnewline
&(0.192)&(0.445)&(0.186)&(0.233)&(0.247)&(0.183)&(0.299) \tabularnewline
(Minus) Unemployment&0.221&0.422&0.154&0.833***&-0.048&0.173&0.269 \tabularnewline
&(0.170)&(0.330)&(0.211)&(0.343)&(0.269)&(0.320)&(0.261) \tabularnewline
Trade&0.249**&0.237*&0.215&0.354*&0.187&0.249&0.205 \tabularnewline
&(0.127)&(0.175)&(0.163)&(0.226)&(0.237)&(0.186)&(0.213) \tabularnewline
\textbf{HDI}&0.197&0.459*&0.048&0.187&0.225&0.228&0.148 \tabularnewline
&(0.167)&(0.285)&(0.204)&(0.253)&(0.311)&(0.272)&(0.240) \tabularnewline
\textbf{Democracy}&0.192**&0.224&0.165&-0.042&0.117&0.050&0.298* \tabularnewline
&(0.101)&(0.259)&(0.131)&(0.142)&(0.151)&(0.060)&(0.194) \tabularnewline
\textbf{General index}&0.276***&0.365***&0.219*&0.217*&0.224*&0.150**&0.332** \tabularnewline
&(0.105)&(0.163)&(0.118)&(0.132)&(0.151)&(0.077)&(0.176) \tabularnewline
\midrule N&2356&653&1703&960&515&739&1399 \tabularnewline
\bottomrule \addlinespace[\belowrulesep]

\end{tabularx}
\\ \parbox{\linewidth}{\footnotesize \caption*{\footnotesize \emph{Notes}: This table reports estimated effects of electoral turnovers for different subsamples. Each estimate corresponds to a separate regression. Column (1) reports results for the full sample. Columns (2) and (3) report results for the subsamples of presidential and parliamentary elections, respectively. Column (4) (resp., 5) shows results on the subsample of elections in which the candidate of the incumbency was the incumbent leader themselves (resp., someone else). Column (6) (resp., 7) shows results for the subsample of elections for which checks and balances on the year before the election were above (resp., below) the median, computed among close elections (i.e., elections with a running variable below 15pp in absolute value). The subsamples of columns (4) and (5) only include elections leading to the designation of a leader in the executive branch. Data Appendix \ref{app:party_leaders} explains how we identify the candidate of the incumbency. Checks and balances are measured as the average of two V-Dem indices: the judicial constraints on the executive index and the legislative constraints on the executive index. Using the method of \cite{clogg1995statistical}, we cannot reject the equality of the estimates for the general index for presidential and parliamentary elections (p-val. = 0.468), for the presence or absence of the incumbent leader on the ballot (p-val. = 0.972), and for high and low checks and balances (p-val. = 0.343). We obtain broadly consistent results when running a parametric regression in which we include the interaction between the treatment and the dimension of heterogeneity. We report the number of observations in the regressions of the table's last row, where the outcome is the general index of country performance. $^{*} p<0.10,^{**} p<0.05,^{***} p<0.01$.}}
\end{table}
